Video Question Answering on MVBench, EgoSchema, LongVideo Bench, and VideoMME
58.5MVBench ScoreAOT
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| AOTPrefilling FLOPs (T)=8.7, FLOPs Ratio=21.3%, Before LLM Retained Ratio=25%2026.03 | 58.5 | 61 | 56 | 56.9 | 58.1 | 99.5 | |
| AOT w DynPrefilling FLOPs (T)=8.7, FLOPs Ratio=21.3%, Before LLM Retained Ratio=25%2026.03 | 58.5 | 61.7 | 56.3 | 57.5 | 58.5 | 100.2 | |
| LLaVA-OV-7BPrefilling FLOPs (T)=40.8, FLOPs Ratio=100%, Before LLM Retained Ratio=100%2026.03 | 58.3 | 60.4 | 56.4 | 58.6 | 58.4 | 100 | |
| AOTPrefilling FLOPs (T)=7.0, FLOPs Ratio=17.2%, Before LLM Retained Ratio=20%2026.03 | 58.3 | 61.4 | 56.1 | 56.8 | 58.2 | 99.7 | |
| AOT w DynPrefilling FLOPs (T)=7.0, FLOPs Ratio=17.2%, Before LLM Retained Ratio=20%2026.03 | 58.2 | 61.5 | 55.9 | 56.8 | 58.1 | 99.5 | |
| VisionZipPrefilling FLOPs (T)=8.7, FLOPs Ratio=21.3%, Before LLM Retained Ratio=25%2026.03 | 57.9 | 60.3 | 56.5 | 58.2 | 58.2 | 99.7 | |
| AOT w DynPrefilling FLOPs (T)=5.2, FLOPs Ratio=12.7%, Before LLM Retained Ratio=15%2026.03 | 57.8 | 61.2 | 55.1 | 56.1 | 57.6 | 98.6 | |
| VisionZipPrefilling FLOPs (T)=7.0, FLOPs Ratio=17.2%, Before LLM Retained Ratio=20%2026.03 | 57.7 | 59.8 | 55.2 | 57.9 | 57.7 | 98.8 | |
| AOTPrefilling FLOPs (T)=5.2, FLOPs Ratio=12.7%, Before LLM Retained Ratio=15%2026.03 | 57.7 | 61.3 | 55.1 | 56.1 | 57.6 | 98.6 | |
| PruneVidPrefilling FLOPs (T)=8.7, FLOPs Ratio=21.3%, Before LLM Retained Ratio=25%2026.03 | 57.4 | 59.9 | 55.7 | 57.4 | 57.6 | 98.6 | |
| PruneVidPrefilling FLOPs (T)=7.0, FLOPs Ratio=17.2%, Before LLM Retained Ratio=20%2026.03 | 57.2 | 59.7 | 54.7 | 56.9 | 57.1 | 97.8 | |
| AOTPrefilling FLOPs (T)=3.4, FLOPs Ratio=8.3%, Before LLM Retained Ratio=10%2026.03 | 57.2 | 60.3 | 53.8 | 56.6 | 57 | 97.6 | |
| AOT w DynPrefilling FLOPs (T)=3.4, FLOPs Ratio=8.3%, Before LLM Retained Ratio=10%2026.03 | 57.2 | 60 | 53.1 | 55.7 | 56.5 | 96.7 | |
| PruneVidPrefilling FLOPs (T)=5.2, FLOPs Ratio=12.7%, Before LLM Retained Ratio=15%2026.03 | 56.8 | 59.7 | 55.4 | 56.6 | 57.1 | 97.8 | |
| FastVIDPrefilling FLOPs (T)=8.7, FLOPs Ratio=21.3%, Before LLM Retained Ratio=25%2026.03 | 56.5 | — | 56.3 | 58 | — | — | |
| VisionZipPrefilling FLOPs (T)=5.2, FLOPs Ratio=12.7%, Before LLM Retained Ratio=15%2026.03 | 56.5 | 59.8 | 54.4 | 56.1 | 56.7 | 97.1 | |
| FastVIDPrefilling FLOPs (T)=7.0, FLOPs Ratio=17.2%, Before LLM Retained Ratio=20%2026.03 | 56.3 | — | 57.1 | 57.9 | — | — | |
| PruneVidPrefilling FLOPs (T)=3.4, FLOPs Ratio=8.3%, Before LLM Retained Ratio=10%2026.03 | 56.2 | 59.8 | 54.5 | 56 | 56.6 | 96.9 | |
| PDropPrefilling FLOPs (T)=10.5, FLOPs Ratio=25.7%, Before LLM Retained Ratio=100%2026.03 | 56.1 | 58 | 54.1 | 56.4 | 56.2 | 96.2 | |
| FastVIDPrefilling FLOPs (T)=5.2, FLOPs Ratio=12.7%, Before LLM Retained Ratio=15%2026.03 | 56 | — | 56.2 | 57.7 | — | — | |
| FastVPrefilling FLOPs (T)=9.3, FLOPs Ratio=22.8%, Before LLM Retained Ratio=100%2026.03 | 55.9 | 57.5 | 56.7 | 56.1 | 56.5 | 96.7 | |
| FastVIDPrefilling FLOPs (T)=3.4, FLOPs Ratio=8.3%, Before LLM Retained Ratio=10%2026.03 | 55.9 | — | 56.3 | 57.3 | — | — | |
| VisionZipPrefilling FLOPs (T)=3.4, FLOPs Ratio=8.3%, Before LLM Retained Ratio=10%2026.03 | 53.5 | 58 | 49.3 | 53.4 | 53.5 | 91.6 | |
| DyCokePrefilling FLOPs (T)=8.7, FLOPs Ratio=21.3%, Before LLM Retained Ratio=25%2026.03 | 53.1 | 59.5 | 49.5 | 54.3 | 54.1 | 92.6 |